Commit | Line | Data |
2605996a |
1 | =head1 NAME |
2 | |
3 | perlthrtut - tutorial on threads in Perl |
4 | |
5 | =head1 DESCRIPTION |
6 | |
53d7eaa8 |
7 | B<NOTE>: this tutorial describes the new Perl threading flavour |
9316ed2f |
8 | introduced in Perl 5.6.0 called interpreter threads, or B<ithreads> |
9 | for short. In this model each thread runs in its own Perl interpreter, |
10 | and any data sharing between threads must be explicit. |
11 | |
12 | There is another older Perl threading flavour called the 5.005 model, |
13 | unsurprisingly for 5.005 versions of Perl. The old model is known to |
14 | have problems, deprecated, and will probably be removed around release |
15 | 5.10. You are strongly encouraged to migrate any existing 5.005 |
16 | threads code to the new model as soon as possible. |
2a4bf773 |
17 | |
53d7eaa8 |
18 | You can see which (or neither) threading flavour you have by |
6eded8f3 |
19 | running C<perl -V> and looking at the C<Platform> section. |
53d7eaa8 |
20 | If you have C<useithreads=define> you have ithreads, if you |
21 | have C<use5005threads=define> you have 5.005 threads. |
22 | If you have neither, you don't have any thread support built in. |
23 | If you have both, you are in trouble. |
2605996a |
24 | |
bfce6503 |
25 | The user-level interface to the 5.005 threads was via the L<Threads> |
26 | class, while ithreads uses the L<threads> class. Note the change in case. |
27 | |
28 | =head1 Status |
29 | |
30 | The ithreads code has been available since Perl 5.6.0, and is considered |
31 | stable. The user-level interface to ithreads (the L<threads> classes) |
32 | appeared in the 5.8.0 release, and as of this time is considered stable, |
33 | although as with all new features, should be treated with caution. |
2605996a |
34 | |
c975c451 |
35 | =head1 What Is A Thread Anyway? |
36 | |
37 | A thread is a flow of control through a program with a single |
38 | execution point. |
39 | |
40 | Sounds an awful lot like a process, doesn't it? Well, it should. |
41 | Threads are one of the pieces of a process. Every process has at least |
42 | one thread and, up until now, every process running Perl had only one |
43 | thread. With 5.8, though, you can create extra threads. We're going |
44 | to show you how, when, and why. |
45 | |
46 | =head1 Threaded Program Models |
47 | |
48 | There are three basic ways that you can structure a threaded |
49 | program. Which model you choose depends on what you need your program |
50 | to do. For many non-trivial threaded programs you'll need to choose |
51 | different models for different pieces of your program. |
52 | |
53 | =head2 Boss/Worker |
54 | |
55 | The boss/worker model usually has one `boss' thread and one or more |
56 | `worker' threads. The boss thread gathers or generates tasks that need |
57 | to be done, then parcels those tasks out to the appropriate worker |
58 | thread. |
59 | |
60 | This model is common in GUI and server programs, where a main thread |
61 | waits for some event and then passes that event to the appropriate |
62 | worker threads for processing. Once the event has been passed on, the |
63 | boss thread goes back to waiting for another event. |
64 | |
65 | The boss thread does relatively little work. While tasks aren't |
66 | necessarily performed faster than with any other method, it tends to |
67 | have the best user-response times. |
68 | |
69 | =head2 Work Crew |
70 | |
71 | In the work crew model, several threads are created that do |
72 | essentially the same thing to different pieces of data. It closely |
73 | mirrors classical parallel processing and vector processors, where a |
74 | large array of processors do the exact same thing to many pieces of |
75 | data. |
76 | |
77 | This model is particularly useful if the system running the program |
78 | will distribute multiple threads across different processors. It can |
79 | also be useful in ray tracing or rendering engines, where the |
80 | individual threads can pass on interim results to give the user visual |
81 | feedback. |
82 | |
83 | =head2 Pipeline |
84 | |
85 | The pipeline model divides up a task into a series of steps, and |
86 | passes the results of one step on to the thread processing the |
87 | next. Each thread does one thing to each piece of data and passes the |
88 | results to the next thread in line. |
89 | |
90 | This model makes the most sense if you have multiple processors so two |
91 | or more threads will be executing in parallel, though it can often |
92 | make sense in other contexts as well. It tends to keep the individual |
93 | tasks small and simple, as well as allowing some parts of the pipeline |
94 | to block (on I/O or system calls, for example) while other parts keep |
95 | going. If you're running different parts of the pipeline on different |
96 | processors you may also take advantage of the caches on each |
97 | processor. |
98 | |
99 | This model is also handy for a form of recursive programming where, |
100 | rather than having a subroutine call itself, it instead creates |
101 | another thread. Prime and Fibonacci generators both map well to this |
102 | form of the pipeline model. (A version of a prime number generator is |
103 | presented later on.) |
104 | |
105 | =head1 Native threads |
106 | |
107 | There are several different ways to implement threads on a system. How |
108 | threads are implemented depends both on the vendor and, in some cases, |
109 | the version of the operating system. Often the first implementation |
110 | will be relatively simple, but later versions of the OS will be more |
111 | sophisticated. |
112 | |
113 | While the information in this section is useful, it's not necessary, |
114 | so you can skip it if you don't feel up to it. |
115 | |
6eded8f3 |
116 | There are three basic categories of threads: user-mode threads, kernel |
c975c451 |
117 | threads, and multiprocessor kernel threads. |
118 | |
119 | User-mode threads are threads that live entirely within a program and |
120 | its libraries. In this model, the OS knows nothing about threads. As |
121 | far as it's concerned, your process is just a process. |
122 | |
123 | This is the easiest way to implement threads, and the way most OSes |
124 | start. The big disadvantage is that, since the OS knows nothing about |
125 | threads, if one thread blocks they all do. Typical blocking activities |
126 | include most system calls, most I/O, and things like sleep(). |
127 | |
128 | Kernel threads are the next step in thread evolution. The OS knows |
129 | about kernel threads, and makes allowances for them. The main |
130 | difference between a kernel thread and a user-mode thread is |
131 | blocking. With kernel threads, things that block a single thread don't |
132 | block other threads. This is not the case with user-mode threads, |
133 | where the kernel blocks at the process level and not the thread level. |
134 | |
135 | This is a big step forward, and can give a threaded program quite a |
136 | performance boost over non-threaded programs. Threads that block |
137 | performing I/O, for example, won't block threads that are doing other |
138 | things. Each process still has only one thread running at once, |
139 | though, regardless of how many CPUs a system might have. |
140 | |
141 | Since kernel threading can interrupt a thread at any time, they will |
142 | uncover some of the implicit locking assumptions you may make in your |
143 | program. For example, something as simple as C<$a = $a + 2> can behave |
144 | unpredictably with kernel threads if $a is visible to other |
145 | threads, as another thread may have changed $a between the time it |
146 | was fetched on the right hand side and the time the new value is |
147 | stored. |
148 | |
6eded8f3 |
149 | Multiprocessor kernel threads are the final step in thread |
c975c451 |
150 | support. With multiprocessor kernel threads on a machine with multiple |
151 | CPUs, the OS may schedule two or more threads to run simultaneously on |
152 | different CPUs. |
153 | |
154 | This can give a serious performance boost to your threaded program, |
155 | since more than one thread will be executing at the same time. As a |
156 | tradeoff, though, any of those nagging synchronization issues that |
157 | might not have shown with basic kernel threads will appear with a |
158 | vengeance. |
159 | |
160 | In addition to the different levels of OS involvement in threads, |
161 | different OSes (and different thread implementations for a particular |
162 | OS) allocate CPU cycles to threads in different ways. |
163 | |
164 | Cooperative multitasking systems have running threads give up control |
165 | if one of two things happen. If a thread calls a yield function, it |
166 | gives up control. It also gives up control if the thread does |
167 | something that would cause it to block, such as perform I/O. In a |
168 | cooperative multitasking implementation, one thread can starve all the |
169 | others for CPU time if it so chooses. |
170 | |
171 | Preemptive multitasking systems interrupt threads at regular intervals |
172 | while the system decides which thread should run next. In a preemptive |
173 | multitasking system, one thread usually won't monopolize the CPU. |
174 | |
175 | On some systems, there can be cooperative and preemptive threads |
176 | running simultaneously. (Threads running with realtime priorities |
177 | often behave cooperatively, for example, while threads running at |
178 | normal priorities behave preemptively.) |
179 | |
bfce6503 |
180 | =head1 What kind of threads are Perl threads? |
c975c451 |
181 | |
182 | If you have experience with other thread implementations, you might |
183 | find that things aren't quite what you expect. It's very important to |
184 | remember when dealing with Perl threads that Perl Threads Are Not X |
185 | Threads, for all values of X. They aren't POSIX threads, or |
186 | DecThreads, or Java's Green threads, or Win32 threads. There are |
187 | similarities, and the broad concepts are the same, but if you start |
188 | looking for implementation details you're going to be either |
189 | disappointed or confused. Possibly both. |
190 | |
191 | This is not to say that Perl threads are completely different from |
192 | everything that's ever come before--they're not. Perl's threading |
193 | model owes a lot to other thread models, especially POSIX. Just as |
194 | Perl is not C, though, Perl threads are not POSIX threads. So if you |
195 | find yourself looking for mutexes, or thread priorities, it's time to |
196 | step back a bit and think about what you want to do and how Perl can |
197 | do it. |
198 | |
6eded8f3 |
199 | However it is important to remember that Perl threads cannot magically |
c975c451 |
200 | do things unless your operating systems threads allows it. So if your |
bfce6503 |
201 | system blocks the entire process on sleep(), Perl usually will as well. |
c975c451 |
202 | |
9316ed2f |
203 | Perl Threads Are Different. |
204 | |
c975c451 |
205 | =head1 Threadsafe Modules |
206 | |
207 | The addition of threads has changed Perl's internals |
208 | substantially. There are implications for people who write |
6eded8f3 |
209 | modules with XS code or external libraries. However, since the threads |
210 | do not share data, pure Perl modules that don't interact with external |
c975c451 |
211 | systems should be safe. Modules that are not tagged as thread-safe should |
212 | be tested or code reviewed before being used in production code. |
213 | |
214 | Not all modules that you might use are thread-safe, and you should |
215 | always assume a module is unsafe unless the documentation says |
216 | otherwise. This includes modules that are distributed as part of the |
217 | core. Threads are a new feature, and even some of the standard |
bfce6503 |
218 | modules aren't thread-safe. |
c975c451 |
219 | |
6eded8f3 |
220 | Even if a module is threadsafe, it doesn't mean that the module is optimized |
221 | to work well with threads. A module could possibly be rewritten to utilize |
222 | the new features in threaded Perl to increase performance in a threaded |
223 | environment. |
c975c451 |
224 | |
225 | If you're using a module that's not thread-safe for some reason, you |
226 | can protect yourself by using semaphores and lots of programming |
227 | discipline to control access to the module. Semaphores are covered |
9316ed2f |
228 | later in the article. |
229 | |
230 | See also L</"Threadsafety of System Libraries">. |
c975c451 |
231 | |
232 | =head1 Thread Basics |
233 | |
234 | The core L<threads> module provides the basic functions you need to write |
235 | threaded programs. In the following sections we'll cover the basics, |
236 | showing you what you need to do to create a threaded program. After |
237 | that, we'll go over some of the features of the L<threads> module that |
238 | make threaded programming easier. |
239 | |
240 | =head2 Basic Thread Support |
241 | |
6eded8f3 |
242 | Thread support is a Perl compile-time option - it's something that's |
c975c451 |
243 | turned on or off when Perl is built at your site, rather than when |
244 | your programs are compiled. If your Perl wasn't compiled with thread |
245 | support enabled, then any attempt to use threads will fail. |
246 | |
c975c451 |
247 | Your programs can use the Config module to check whether threads are |
248 | enabled. If your program can't run without them, you can say something |
249 | like: |
250 | |
9316ed2f |
251 | $Config{useithreads} or die "Recompile Perl with threads to run this program."; |
c975c451 |
252 | |
253 | A possibly-threaded program using a possibly-threaded module might |
254 | have code like this: |
255 | |
256 | use Config; |
257 | use MyMod; |
258 | |
9316ed2f |
259 | BEGIN { |
260 | if ($Config{useithreads}) { |
261 | # We have threads |
262 | require MyMod_threaded; |
263 | import MyMod_threaded; |
264 | } else { |
265 | require MyMod_unthreaded; |
266 | import MyMod_unthreaded; |
267 | } |
c975c451 |
268 | } |
269 | |
270 | Since code that runs both with and without threads is usually pretty |
271 | messy, it's best to isolate the thread-specific code in its own |
272 | module. In our example above, that's what MyMod_threaded is, and it's |
273 | only imported if we're running on a threaded Perl. |
274 | |
8f95bfb9 |
275 | =head2 A Note about the Examples |
276 | |
277 | Although thread support is considered to be stable, there are still a number |
278 | of quirks that may startle you when you try out any of the examples below. |
279 | In a real situation, care should be taken that all threads are finished |
280 | executing before the program exits. That care has B<not> been taken in these |
281 | examples in the interest of simplicity. Running these examples "as is" will |
282 | produce error messages, usually caused by the fact that there are still |
283 | threads running when the program exits. You should not be alarmed by this. |
284 | Future versions of Perl may fix this problem. |
285 | |
c975c451 |
286 | =head2 Creating Threads |
287 | |
288 | The L<threads> package provides the tools you need to create new |
289 | threads. Like any other module, you need to tell Perl you want to use |
290 | it; C<use threads> imports all the pieces you need to create basic |
291 | threads. |
292 | |
293 | The simplest, straightforward way to create a thread is with new(): |
294 | |
295 | use threads; |
296 | |
297 | $thr = threads->new(\&sub1); |
298 | |
299 | sub sub1 { |
300 | print "In the thread\n"; |
301 | } |
302 | |
303 | The new() method takes a reference to a subroutine and creates a new |
304 | thread, which starts executing in the referenced subroutine. Control |
305 | then passes both to the subroutine and the caller. |
306 | |
307 | If you need to, your program can pass parameters to the subroutine as |
308 | part of the thread startup. Just include the list of parameters as |
309 | part of the C<threads::new> call, like this: |
310 | |
311 | use threads; |
bfce6503 |
312 | |
c975c451 |
313 | $Param3 = "foo"; |
314 | $thr = threads->new(\&sub1, "Param 1", "Param 2", $Param3); |
315 | $thr = threads->new(\&sub1, @ParamList); |
8f95bfb9 |
316 | $thr = threads->new(\&sub1, qw(Param1 Param2 Param3)); |
c975c451 |
317 | |
318 | sub sub1 { |
319 | my @InboundParameters = @_; |
320 | print "In the thread\n"; |
321 | print "got parameters >", join("<>", @InboundParameters), "<\n"; |
322 | } |
323 | |
324 | |
325 | The last example illustrates another feature of threads. You can spawn |
326 | off several threads using the same subroutine. Each thread executes |
327 | the same subroutine, but in a separate thread with a separate |
328 | environment and potentially separate arguments. |
329 | |
9316ed2f |
330 | C<create()> is a synonym for C<new()>. |
bfce6503 |
331 | |
c975c451 |
332 | =head2 Giving up control |
333 | |
334 | There are times when you may find it useful to have a thread |
335 | explicitly give up the CPU to another thread. Your threading package |
336 | might not support preemptive multitasking for threads, for example, or |
337 | you may be doing something compute-intensive and want to make sure |
338 | that the user-interface thread gets called frequently. Regardless, |
339 | there are times that you might want a thread to give up the processor. |
340 | |
341 | Perl's threading package provides the yield() function that does |
342 | this. yield() is pretty straightforward, and works like this: |
343 | |
344 | use threads; |
818c4caa |
345 | |
bfce6503 |
346 | sub loop { |
347 | my $thread = shift; |
348 | my $foo = 50; |
349 | while($foo--) { print "in thread $thread\n" } |
8f95bfb9 |
350 | threads->yield; |
bfce6503 |
351 | $foo = 50; |
352 | while($foo--) { print "in thread $thread\n" } |
353 | } |
c975c451 |
354 | |
bfce6503 |
355 | my $thread1 = threads->new(\&loop, 'first'); |
356 | my $thread2 = threads->new(\&loop, 'second'); |
357 | my $thread3 = threads->new(\&loop, 'third'); |
818c4caa |
358 | |
c975c451 |
359 | It is important to remember that yield() is only a hint to give up the CPU, |
360 | it depends on your hardware, OS and threading libraries what actually happens. |
361 | Therefore it is important to note that one should not build the scheduling of |
362 | the threads around yield() calls. It might work on your platform but it won't |
363 | work on another platform. |
364 | |
365 | =head2 Waiting For A Thread To Exit |
366 | |
367 | Since threads are also subroutines, they can return values. To wait |
6eded8f3 |
368 | for a thread to exit and extract any values it might return, you can |
369 | use the join() method: |
c975c451 |
370 | |
371 | use threads; |
bfce6503 |
372 | |
c975c451 |
373 | $thr = threads->new(\&sub1); |
374 | |
375 | @ReturnData = $thr->join; |
376 | print "Thread returned @ReturnData"; |
377 | |
378 | sub sub1 { return "Fifty-six", "foo", 2; } |
379 | |
380 | In the example above, the join() method returns as soon as the thread |
381 | ends. In addition to waiting for a thread to finish and gathering up |
382 | any values that the thread might have returned, join() also performs |
383 | any OS cleanup necessary for the thread. That cleanup might be |
384 | important, especially for long-running programs that spawn lots of |
385 | threads. If you don't want the return values and don't want to wait |
386 | for the thread to finish, you should call the detach() method |
bfce6503 |
387 | instead, as described next. |
c975c451 |
388 | |
389 | =head2 Ignoring A Thread |
390 | |
391 | join() does three things: it waits for a thread to exit, cleans up |
392 | after it, and returns any data the thread may have produced. But what |
393 | if you're not interested in the thread's return values, and you don't |
394 | really care when the thread finishes? All you want is for the thread |
395 | to get cleaned up after when it's done. |
396 | |
397 | In this case, you use the detach() method. Once a thread is detached, |
398 | it'll run until it's finished, then Perl will clean up after it |
399 | automatically. |
400 | |
401 | use threads; |
bfce6503 |
402 | |
6eded8f3 |
403 | $thr = threads->new(\&sub1); # Spawn the thread |
c975c451 |
404 | |
405 | $thr->detach; # Now we officially don't care any more |
406 | |
407 | sub sub1 { |
408 | $a = 0; |
409 | while (1) { |
410 | $a++; |
411 | print "\$a is $a\n"; |
412 | sleep 1; |
413 | } |
414 | } |
415 | |
bfce6503 |
416 | Once a thread is detached, it may not be joined, and any return data |
417 | that it might have produced (if it was done and waiting for a join) is |
c975c451 |
418 | lost. |
419 | |
420 | =head1 Threads And Data |
421 | |
422 | Now that we've covered the basics of threads, it's time for our next |
423 | topic: data. Threading introduces a couple of complications to data |
424 | access that non-threaded programs never need to worry about. |
425 | |
426 | =head2 Shared And Unshared Data |
427 | |
bfce6503 |
428 | The biggest difference between Perl ithreads and the old 5.005 style |
429 | threading, or for that matter, to most other threading systems out there, |
430 | is that by default, no data is shared. When a new perl thread is created, |
431 | all the data associated with the current thread is copied to the new |
432 | thread, and is subsequently private to that new thread! |
433 | This is similar in feel to what happens when a UNIX process forks, |
434 | except that in this case, the data is just copied to a different part of |
435 | memory within the same process rather than a real fork taking place. |
c975c451 |
436 | |
437 | To make use of threading however, one usually want the threads to share |
bfce6503 |
438 | at least some data between themselves. This is done with the |
439 | L<threads::shared> module and the C< : shared> attribute: |
440 | |
441 | use threads; |
442 | use threads::shared; |
443 | |
444 | my $foo : shared = 1; |
445 | my $bar = 1; |
446 | threads->new(sub { $foo++; $bar++ })->join; |
818c4caa |
447 | |
bfce6503 |
448 | print "$foo\n"; #prints 2 since $foo is shared |
449 | print "$bar\n"; #prints 1 since $bar is not shared |
450 | |
451 | In the case of a shared array, all the array's elements are shared, and for |
452 | a shared hash, all the keys and values are shared. This places |
453 | restrictions on what may be assigned to shared array and hash elements: only |
454 | simple values or references to shared variables are allowed - this is |
f3278b06 |
455 | so that a private variable can't accidentally become shared. A bad |
bfce6503 |
456 | assignment will cause the thread to die. For example: |
457 | |
458 | use threads; |
459 | use threads::shared; |
460 | |
461 | my $var = 1; |
462 | my $svar : shared = 2; |
463 | my %hash : shared; |
464 | |
465 | ... create some threads ... |
466 | |
467 | $hash{a} = 1; # all threads see exists($hash{a}) and $hash{a} == 1 |
468 | $hash{a} = $var # okay - copy-by-value: same affect as previous |
469 | $hash{a} = $svar # okay - copy-by-value: same affect as previous |
470 | $hash{a} = \$svar # okay - a reference to a shared variable |
471 | $hash{a} = \$var # This will die |
472 | delete $hash{a} # okay - all threads will see !exists($hash{a}) |
473 | |
474 | Note that a shared variable guarantees that if two or more threads try to |
475 | modify it at the same time, the internal state of the variable will not |
476 | become corrupted. However, there are no guarantees beyond this, as |
477 | explained in the next section. |
c975c451 |
478 | |
6eded8f3 |
479 | =head2 Thread Pitfalls: Races |
c975c451 |
480 | |
481 | While threads bring a new set of useful tools, they also bring a |
482 | number of pitfalls. One pitfall is the race condition: |
483 | |
484 | use threads; |
485 | use threads::shared; |
bfce6503 |
486 | |
c975c451 |
487 | my $a : shared = 1; |
488 | $thr1 = threads->new(\&sub1); |
489 | $thr2 = threads->new(\&sub2); |
490 | |
491 | $thr1->join; |
492 | $thr2->join; |
493 | print "$a\n"; |
494 | |
bfce6503 |
495 | sub sub1 { my $foo = $a; $a = $foo + 1; } |
496 | sub sub2 { my $bar = $a; $a = $bar + 1; } |
c975c451 |
497 | |
498 | What do you think $a will be? The answer, unfortunately, is "it |
499 | depends." Both sub1() and sub2() access the global variable $a, once |
500 | to read and once to write. Depending on factors ranging from your |
501 | thread implementation's scheduling algorithm to the phase of the moon, |
502 | $a can be 2 or 3. |
503 | |
504 | Race conditions are caused by unsynchronized access to shared |
505 | data. Without explicit synchronization, there's no way to be sure that |
506 | nothing has happened to the shared data between the time you access it |
507 | and the time you update it. Even this simple code fragment has the |
508 | possibility of error: |
509 | |
510 | use threads; |
511 | my $a : shared = 2; |
512 | my $b : shared; |
513 | my $c : shared; |
514 | my $thr1 = threads->create(sub { $b = $a; $a = $b + 1; }); |
515 | my $thr2 = threads->create(sub { $c = $a; $a = $c + 1; }); |
8f95bfb9 |
516 | $thr1->join; |
517 | $thr2->join; |
c975c451 |
518 | |
519 | Two threads both access $a. Each thread can potentially be interrupted |
520 | at any point, or be executed in any order. At the end, $a could be 3 |
521 | or 4, and both $b and $c could be 2 or 3. |
522 | |
bfce6503 |
523 | Even C<$a += 5> or C<$a++> are not guaranteed to be atomic. |
524 | |
c975c451 |
525 | Whenever your program accesses data or resources that can be accessed |
526 | by other threads, you must take steps to coordinate access or risk |
bfce6503 |
527 | data inconsistency and race conditions. Note that Perl will protect its |
528 | internals from your race conditions, but it won't protect you from you. |
529 | |
f3278b06 |
530 | =head1 Synchronization and control |
bfce6503 |
531 | |
532 | Perl provides a number of mechanisms to coordinate the interactions |
533 | between themselves and their data, to avoid race conditions and the like. |
534 | Some of these are designed to resemble the common techniques used in thread |
535 | libraries such as C<pthreads>; others are Perl-specific. Often, the |
f3278b06 |
536 | standard techniques are clumsily and difficult to get right (such as |
bfce6503 |
537 | condition waits). Where possible, it is usually easier to use Perlish |
538 | techniques such as queues, which remove some of the hard work involved. |
c975c451 |
539 | |
540 | =head2 Controlling access: lock() |
541 | |
542 | The lock() function takes a shared variable and puts a lock on it. |
bfce6503 |
543 | No other thread may lock the variable until the the variable is unlocked |
544 | by the thread holding the lock. Unlocking happens automatically |
8f95bfb9 |
545 | when the locking thread exits the outermost block that contains |
bfce6503 |
546 | C<lock()> function. Using lock() is straightforward: this example has |
f3278b06 |
547 | several threads doing some calculations in parallel, and occasionally |
bfce6503 |
548 | updating a running total: |
549 | |
550 | use threads; |
551 | use threads::shared; |
552 | |
553 | my $total : shared = 0; |
554 | |
555 | sub calc { |
556 | for (;;) { |
557 | my $result; |
558 | # (... do some calculations and set $result ...) |
559 | { |
560 | lock($total); # block until we obtain the lock |
8f95bfb9 |
561 | $total += $result; |
f3278b06 |
562 | } # lock implicitly released at end of scope |
bfce6503 |
563 | last if $result == 0; |
564 | } |
565 | } |
566 | |
567 | my $thr1 = threads->new(\&calc); |
568 | my $thr2 = threads->new(\&calc); |
569 | my $thr3 = threads->new(\&calc); |
570 | $thr1->join; |
571 | $thr2->join; |
572 | $thr3->join; |
573 | print "total=$total\n"; |
c975c451 |
574 | |
c975c451 |
575 | |
576 | lock() blocks the thread until the variable being locked is |
577 | available. When lock() returns, your thread can be sure that no other |
bfce6503 |
578 | thread can lock that variable until the outermost block containing the |
c975c451 |
579 | lock exits. |
580 | |
581 | It's important to note that locks don't prevent access to the variable |
582 | in question, only lock attempts. This is in keeping with Perl's |
583 | longstanding tradition of courteous programming, and the advisory file |
584 | locking that flock() gives you. |
585 | |
586 | You may lock arrays and hashes as well as scalars. Locking an array, |
587 | though, will not block subsequent locks on array elements, just lock |
588 | attempts on the array itself. |
589 | |
bfce6503 |
590 | Locks are recursive, which means it's okay for a thread to |
c975c451 |
591 | lock a variable more than once. The lock will last until the outermost |
bfce6503 |
592 | lock() on the variable goes out of scope. For example: |
593 | |
594 | my $x : shared; |
595 | doit(); |
596 | |
597 | sub doit { |
598 | { |
599 | { |
600 | lock($x); # wait for lock |
8f95bfb9 |
601 | lock($x); # NOOP - we already have the lock |
bfce6503 |
602 | { |
603 | lock($x); # NOOP |
604 | { |
605 | lock($x); # NOOP |
606 | lockit_some_more(); |
607 | } |
608 | } |
609 | } # *** implicit unlock here *** |
610 | } |
611 | } |
612 | |
613 | sub lockit_some_more { |
614 | lock($x); # NOOP |
615 | } # nothing happens here |
616 | |
617 | Note that there is no unlock() function - the only way to unlock a |
618 | variable is to allow it to go out of scope. |
619 | |
620 | A lock can either be used to guard the data contained within the variable |
621 | being locked, or it can be used to guard something else, like a section |
622 | of code. In this latter case, the variable in question does not hold any |
623 | useful data, and exists only for the purpose of being locked. In this |
624 | respect, the variable behaves like the mutexes and basic semaphores of |
625 | traditional thread libraries. |
c975c451 |
626 | |
bfce6503 |
627 | =head2 A Thread Pitfall: Deadlocks |
c975c451 |
628 | |
bfce6503 |
629 | Locks are a handy tool to synchronize access to data, and using them |
c975c451 |
630 | properly is the key to safe shared data. Unfortunately, locks aren't |
f3278b06 |
631 | without their dangers, especially when multiple locks are involved. |
bfce6503 |
632 | Consider the following code: |
c975c451 |
633 | |
634 | use threads; |
bfce6503 |
635 | |
c975c451 |
636 | my $a : shared = 4; |
637 | my $b : shared = "foo"; |
638 | my $thr1 = threads->new(sub { |
639 | lock($a); |
bfce6503 |
640 | threads->yield; |
c975c451 |
641 | sleep 20; |
bfce6503 |
642 | lock($b); |
c975c451 |
643 | }); |
644 | my $thr2 = threads->new(sub { |
645 | lock($b); |
bfce6503 |
646 | threads->yield; |
c975c451 |
647 | sleep 20; |
bfce6503 |
648 | lock($a); |
c975c451 |
649 | }); |
650 | |
651 | This program will probably hang until you kill it. The only way it |
bfce6503 |
652 | won't hang is if one of the two threads acquires both locks |
c975c451 |
653 | first. A guaranteed-to-hang version is more complicated, but the |
654 | principle is the same. |
655 | |
bfce6503 |
656 | The first thread will grab a lock on $a, then, after a pause during which |
657 | the second thread has probably had time to do some work, try to grab a |
658 | lock on $b. Meanwhile, the second thread grabs a lock on $b, then later |
659 | tries to grab a lock on $a. The second lock attempt for both threads will |
660 | block, each waiting for the other to release its lock. |
c975c451 |
661 | |
662 | This condition is called a deadlock, and it occurs whenever two or |
663 | more threads are trying to get locks on resources that the others |
664 | own. Each thread will block, waiting for the other to release a lock |
665 | on a resource. That never happens, though, since the thread with the |
666 | resource is itself waiting for a lock to be released. |
667 | |
668 | There are a number of ways to handle this sort of problem. The best |
669 | way is to always have all threads acquire locks in the exact same |
670 | order. If, for example, you lock variables $a, $b, and $c, always lock |
671 | $a before $b, and $b before $c. It's also best to hold on to locks for |
672 | as short a period of time to minimize the risks of deadlock. |
673 | |
48b96218 |
674 | The other synchronization primitives described below can suffer from |
bfce6503 |
675 | similar problems. |
676 | |
c975c451 |
677 | =head2 Queues: Passing Data Around |
678 | |
679 | A queue is a special thread-safe object that lets you put data in one |
680 | end and take it out the other without having to worry about |
681 | synchronization issues. They're pretty straightforward, and look like |
682 | this: |
683 | |
684 | use threads; |
685 | use threads::shared::queue; |
686 | |
8f95bfb9 |
687 | my $DataQueue = threads::shared::queue->new; |
c975c451 |
688 | $thr = threads->new(sub { |
689 | while ($DataElement = $DataQueue->dequeue) { |
690 | print "Popped $DataElement off the queue\n"; |
691 | } |
692 | }); |
693 | |
694 | $DataQueue->enqueue(12); |
695 | $DataQueue->enqueue("A", "B", "C"); |
696 | $DataQueue->enqueue(\$thr); |
697 | sleep 10; |
698 | $DataQueue->enqueue(undef); |
8f95bfb9 |
699 | $thr->join; |
c975c451 |
700 | |
6eded8f3 |
701 | You create the queue with C<new threads::shared::queue>. Then you can |
702 | add lists of scalars onto the end with enqueue(), and pop scalars off |
703 | the front of it with dequeue(). A queue has no fixed size, and can grow |
704 | as needed to hold everything pushed on to it. |
c975c451 |
705 | |
706 | If a queue is empty, dequeue() blocks until another thread enqueues |
707 | something. This makes queues ideal for event loops and other |
708 | communications between threads. |
709 | |
c975c451 |
710 | =head2 Semaphores: Synchronizing Data Access |
711 | |
bfce6503 |
712 | Semaphores are a kind of generic locking mechanism. In their most basic |
713 | form, they behave very much like lockable scalars, except that thay |
714 | can't hold data, and that they must be explicitly unlocked. In their |
715 | advanced form, they act like a kind of counter, and can allow multiple |
716 | threads to have the 'lock' at any one time. |
2605996a |
717 | |
bfce6503 |
718 | =head2 Basic semaphores |
2605996a |
719 | |
bfce6503 |
720 | Semaphores have two methods, down() and up(): down() decrements the resource |
721 | count, while up increments it. Calls to down() will block if the |
c975c451 |
722 | semaphore's current count would decrement below zero. This program |
723 | gives a quick demonstration: |
724 | |
725 | use threads qw(yield); |
726 | use threads::shared::semaphore; |
bfce6503 |
727 | |
c975c451 |
728 | my $semaphore = new threads::shared::semaphore; |
bfce6503 |
729 | my $GlobalVariable : shared = 0; |
2605996a |
730 | |
c975c451 |
731 | $thr1 = new threads \&sample_sub, 1; |
732 | $thr2 = new threads \&sample_sub, 2; |
733 | $thr3 = new threads \&sample_sub, 3; |
2605996a |
734 | |
c975c451 |
735 | sub sample_sub { |
736 | my $SubNumber = shift @_; |
737 | my $TryCount = 10; |
738 | my $LocalCopy; |
739 | sleep 1; |
740 | while ($TryCount--) { |
741 | $semaphore->down; |
742 | $LocalCopy = $GlobalVariable; |
743 | print "$TryCount tries left for sub $SubNumber (\$GlobalVariable is $GlobalVariable)\n"; |
744 | yield; |
745 | sleep 2; |
746 | $LocalCopy++; |
747 | $GlobalVariable = $LocalCopy; |
748 | $semaphore->up; |
749 | } |
750 | } |
6eded8f3 |
751 | |
8f95bfb9 |
752 | $thr1->join; |
753 | $thr2->join; |
754 | $thr3->join; |
2605996a |
755 | |
c975c451 |
756 | The three invocations of the subroutine all operate in sync. The |
757 | semaphore, though, makes sure that only one thread is accessing the |
758 | global variable at once. |
2605996a |
759 | |
bfce6503 |
760 | =head2 Advanced Semaphores |
2605996a |
761 | |
c975c451 |
762 | By default, semaphores behave like locks, letting only one thread |
763 | down() them at a time. However, there are other uses for semaphores. |
2605996a |
764 | |
6eded8f3 |
765 | Each semaphore has a counter attached to it. By default, semaphores are |
766 | created with the counter set to one, down() decrements the counter by |
767 | one, and up() increments by one. However, we can override any or all |
768 | of these defaults simply by passing in different values: |
769 | |
770 | use threads; |
771 | use threads::shared::semaphore; |
772 | my $semaphore = threads::shared::semaphore->new(5); |
773 | # Creates a semaphore with the counter set to five |
774 | |
775 | $thr1 = threads->new(\&sub1); |
776 | $thr2 = threads->new(\&sub1); |
777 | |
778 | sub sub1 { |
779 | $semaphore->down(5); # Decrements the counter by five |
780 | # Do stuff here |
781 | $semaphore->up(5); # Increment the counter by five |
782 | } |
783 | |
8f95bfb9 |
784 | $thr1->detach; |
785 | $thr2->detach; |
6eded8f3 |
786 | |
787 | If down() attempts to decrement the counter below zero, it blocks until |
788 | the counter is large enough. Note that while a semaphore can be created |
789 | with a starting count of zero, any up() or down() always changes the |
790 | counter by at least one, and so $semaphore->down(0) is the same as |
791 | $semaphore->down(1). |
2605996a |
792 | |
c975c451 |
793 | The question, of course, is why would you do something like this? Why |
794 | create a semaphore with a starting count that's not one, or why |
795 | decrement/increment it by more than one? The answer is resource |
796 | availability. Many resources that you want to manage access for can be |
797 | safely used by more than one thread at once. |
2605996a |
798 | |
c975c451 |
799 | For example, let's take a GUI driven program. It has a semaphore that |
800 | it uses to synchronize access to the display, so only one thread is |
801 | ever drawing at once. Handy, but of course you don't want any thread |
802 | to start drawing until things are properly set up. In this case, you |
803 | can create a semaphore with a counter set to zero, and up it when |
804 | things are ready for drawing. |
2605996a |
805 | |
c975c451 |
806 | Semaphores with counters greater than one are also useful for |
807 | establishing quotas. Say, for example, that you have a number of |
808 | threads that can do I/O at once. You don't want all the threads |
809 | reading or writing at once though, since that can potentially swamp |
810 | your I/O channels, or deplete your process' quota of filehandles. You |
811 | can use a semaphore initialized to the number of concurrent I/O |
812 | requests (or open files) that you want at any one time, and have your |
813 | threads quietly block and unblock themselves. |
2605996a |
814 | |
c975c451 |
815 | Larger increments or decrements are handy in those cases where a |
816 | thread needs to check out or return a number of resources at once. |
2605996a |
817 | |
bfce6503 |
818 | =head2 cond_wait() and cond_signal() |
819 | |
820 | These two functions can be used in conjunction with locks to notify |
821 | co-operating threads that a resource has become available. They are |
822 | very similar in use to the functions found in C<pthreads>. However |
823 | for most purposes, queues are simpler to use and more intuitive. See |
824 | L<threads::shared> for more details. |
2605996a |
825 | |
c975c451 |
826 | =head1 General Thread Utility Routines |
827 | |
828 | We've covered the workhorse parts of Perl's threading package, and |
829 | with these tools you should be well on your way to writing threaded |
830 | code and packages. There are a few useful little pieces that didn't |
831 | really fit in anyplace else. |
832 | |
833 | =head2 What Thread Am I In? |
834 | |
bfce6503 |
835 | The C<< threads->self >> class method provides your program with a way to |
836 | get an object representing the thread it's currently in. You can use this |
6eded8f3 |
837 | object in the same way as the ones returned from thread creation. |
c975c451 |
838 | |
839 | =head2 Thread IDs |
840 | |
841 | tid() is a thread object method that returns the thread ID of the |
842 | thread the object represents. Thread IDs are integers, with the main |
843 | thread in a program being 0. Currently Perl assigns a unique tid to |
844 | every thread ever created in your program, assigning the first thread |
845 | to be created a tid of 1, and increasing the tid by 1 for each new |
846 | thread that's created. |
847 | |
848 | =head2 Are These Threads The Same? |
849 | |
850 | The equal() method takes two thread objects and returns true |
851 | if the objects represent the same thread, and false if they don't. |
852 | |
853 | Thread objects also have an overloaded == comparison so that you can do |
854 | comparison on them as you would with normal objects. |
855 | |
856 | =head2 What Threads Are Running? |
857 | |
bfce6503 |
858 | C<< threads->list >> returns a list of thread objects, one for each thread |
c975c451 |
859 | that's currently running and not detached. Handy for a number of things, |
860 | including cleaning up at the end of your program: |
861 | |
862 | # Loop through all the threads |
863 | foreach $thr (threads->list) { |
864 | # Don't join the main thread or ourselves |
865 | if ($thr->tid && !threads::equal($thr, threads->self)) { |
866 | $thr->join; |
867 | } |
868 | } |
869 | |
bfce6503 |
870 | If some threads have not finished running when the main Perl thread |
871 | ends, Perl will warn you about it and die, since it is impossible for Perl |
6eded8f3 |
872 | to clean up itself while other threads are running |
c975c451 |
873 | |
874 | =head1 A Complete Example |
875 | |
876 | Confused yet? It's time for an example program to show some of the |
877 | things we've covered. This program finds prime numbers using threads. |
878 | |
879 | 1 #!/usr/bin/perl -w |
880 | 2 # prime-pthread, courtesy of Tom Christiansen |
881 | 3 |
882 | 4 use strict; |
883 | 5 |
884 | 6 use threads; |
885 | 7 use threads::shared::queue; |
886 | 8 |
887 | 9 my $stream = new threads::shared::queue; |
888 | 10 my $kid = new threads(\&check_num, $stream, 2); |
889 | 11 |
890 | 12 for my $i ( 3 .. 1000 ) { |
891 | 13 $stream->enqueue($i); |
892 | 14 } |
893 | 15 |
894 | 16 $stream->enqueue(undef); |
8f95bfb9 |
895 | 17 $kid->join; |
c975c451 |
896 | 18 |
897 | 19 sub check_num { |
898 | 20 my ($upstream, $cur_prime) = @_; |
899 | 21 my $kid; |
900 | 22 my $downstream = new threads::shared::queue; |
901 | 23 while (my $num = $upstream->dequeue) { |
902 | 24 next unless $num % $cur_prime; |
903 | 25 if ($kid) { |
904 | 26 $downstream->enqueue($num); |
905 | 27 } else { |
906 | 28 print "Found prime $num\n"; |
907 | 29 $kid = new threads(\&check_num, $downstream, $num); |
908 | 30 } |
909 | 31 } |
910 | 32 $downstream->enqueue(undef) if $kid; |
8f95bfb9 |
911 | 33 $kid->join if $kid; |
c975c451 |
912 | 34 } |
913 | |
914 | This program uses the pipeline model to generate prime numbers. Each |
915 | thread in the pipeline has an input queue that feeds numbers to be |
916 | checked, a prime number that it's responsible for, and an output queue |
6eded8f3 |
917 | that into which it funnels numbers that have failed the check. If the thread |
c975c451 |
918 | has a number that's failed its check and there's no child thread, then |
919 | the thread must have found a new prime number. In that case, a new |
920 | child thread is created for that prime and stuck on the end of the |
921 | pipeline. |
922 | |
6eded8f3 |
923 | This probably sounds a bit more confusing than it really is, so let's |
c975c451 |
924 | go through this program piece by piece and see what it does. (For |
925 | those of you who might be trying to remember exactly what a prime |
926 | number is, it's a number that's only evenly divisible by itself and 1) |
927 | |
928 | The bulk of the work is done by the check_num() subroutine, which |
929 | takes a reference to its input queue and a prime number that it's |
930 | responsible for. After pulling in the input queue and the prime that |
931 | the subroutine's checking (line 20), we create a new queue (line 22) |
932 | and reserve a scalar for the thread that we're likely to create later |
933 | (line 21). |
934 | |
935 | The while loop from lines 23 to line 31 grabs a scalar off the input |
936 | queue and checks against the prime this thread is responsible |
937 | for. Line 24 checks to see if there's a remainder when we modulo the |
938 | number to be checked against our prime. If there is one, the number |
939 | must not be evenly divisible by our prime, so we need to either pass |
940 | it on to the next thread if we've created one (line 26) or create a |
941 | new thread if we haven't. |
942 | |
943 | The new thread creation is line 29. We pass on to it a reference to |
944 | the queue we've created, and the prime number we've found. |
945 | |
946 | Finally, once the loop terminates (because we got a 0 or undef in the |
947 | queue, which serves as a note to die), we pass on the notice to our |
6eded8f3 |
948 | child and wait for it to exit if we've created a child (lines 32 and |
c975c451 |
949 | 37). |
950 | |
951 | Meanwhile, back in the main thread, we create a queue (line 9) and the |
952 | initial child thread (line 10), and pre-seed it with the first prime: |
953 | 2. Then we queue all the numbers from 3 to 1000 for checking (lines |
954 | 12-14), then queue a die notice (line 16) and wait for the first child |
955 | thread to terminate (line 17). Because a child won't die until its |
956 | child has died, we know that we're done once we return from the join. |
957 | |
958 | That's how it works. It's pretty simple; as with many Perl programs, |
959 | the explanation is much longer than the program. |
960 | |
bfce6503 |
961 | =head1 Performance considerations |
962 | |
963 | The main thing to bear in mind when comparing ithreads to other threading |
964 | models is the fact that for each new thread created, a complete copy of |
965 | all the variables and data of the parent thread has to be taken. Thus |
966 | thread creation can be quite expensive, both in terms of memory usage and |
967 | time spent in creation. The ideal way to reduce these costs is to have a |
968 | relatively short number of long-lived threads, all created fairly early |
969 | on - before the base thread has accumulated too much data. Of course, this |
970 | may not always be possible, so compromises have to be made. However, after |
971 | a thread has been created, its performance and extra memory usage should |
972 | be little different than ordinary code. |
973 | |
974 | Also note that under the current implementation, shared variables |
975 | use a little more memory and are a little slower than ordinary variables. |
976 | |
bdcfa4c7 |
977 | =head1 Threadsafety of System Libraries |
978 | |
979 | Whether various library calls are threadsafe is outside the control |
80bbcbc4 |
980 | of Perl. Calls often suffering from not being threadsafe include: |
bdcfa4c7 |
981 | localtime(), gmtime(), get{gr,host,net,proto,serv,pw}*(), readdir(), |
80bbcbc4 |
982 | rand(), and srand() -- in general, calls that depend on some external |
983 | state. |
984 | |
985 | If the system Perl is compiled in has threadsafe variants of such |
986 | calls, they will be used. Beyond that, Perl is at the mercy of |
987 | the threadsafety or unsafety of the calls. Please consult your |
988 | C library call documentation. |
989 | |
990 | In some platforms the threadsafe interfaces may fail if the result |
991 | buffer is too small (for example getgrent() may return quite large |
992 | group member lists). Perl will retry growing the result buffer |
993 | a few times, but only up to 64k (for safety reasons). |
bdcfa4c7 |
994 | |
c975c451 |
995 | =head1 Conclusion |
996 | |
997 | A complete thread tutorial could fill a book (and has, many times), |
6eded8f3 |
998 | but with what we've covered in this introduction, you should be well |
999 | on your way to becoming a threaded Perl expert. |
c975c451 |
1000 | |
1001 | =head1 Bibliography |
1002 | |
1003 | Here's a short bibliography courtesy of Jürgen Christoffel: |
1004 | |
1005 | =head2 Introductory Texts |
1006 | |
1007 | Birrell, Andrew D. An Introduction to Programming with |
1008 | Threads. Digital Equipment Corporation, 1989, DEC-SRC Research Report |
1009 | #35 online as |
6eded8f3 |
1010 | http://gatekeeper.dec.com/pub/DEC/SRC/research-reports/abstracts/src-rr-035.html |
1011 | (highly recommended) |
c975c451 |
1012 | |
1013 | Robbins, Kay. A., and Steven Robbins. Practical Unix Programming: A |
1014 | Guide to Concurrency, Communication, and |
1015 | Multithreading. Prentice-Hall, 1996. |
1016 | |
1017 | Lewis, Bill, and Daniel J. Berg. Multithreaded Programming with |
1018 | Pthreads. Prentice Hall, 1997, ISBN 0-13-443698-9 (a well-written |
1019 | introduction to threads). |
1020 | |
1021 | Nelson, Greg (editor). Systems Programming with Modula-3. Prentice |
1022 | Hall, 1991, ISBN 0-13-590464-1. |
1023 | |
1024 | Nichols, Bradford, Dick Buttlar, and Jacqueline Proulx Farrell. |
1025 | Pthreads Programming. O'Reilly & Associates, 1996, ISBN 156592-115-1 |
1026 | (covers POSIX threads). |
1027 | |
1028 | =head2 OS-Related References |
1029 | |
1030 | Boykin, Joseph, David Kirschen, Alan Langerman, and Susan |
1031 | LoVerso. Programming under Mach. Addison-Wesley, 1994, ISBN |
1032 | 0-201-52739-1. |
1033 | |
1034 | Tanenbaum, Andrew S. Distributed Operating Systems. Prentice Hall, |
1035 | 1995, ISBN 0-13-219908-4 (great textbook). |
1036 | |
1037 | Silberschatz, Abraham, and Peter B. Galvin. Operating System Concepts, |
1038 | 4th ed. Addison-Wesley, 1995, ISBN 0-201-59292-4 |
1039 | |
1040 | =head2 Other References |
1041 | |
1042 | Arnold, Ken and James Gosling. The Java Programming Language, 2nd |
1043 | ed. Addison-Wesley, 1998, ISBN 0-201-31006-6. |
1044 | |
1045 | Le Sergent, T. and B. Berthomieu. "Incremental MultiThreaded Garbage |
1046 | Collection on Virtually Shared Memory Architectures" in Memory |
1047 | Management: Proc. of the International Workshop IWMM 92, St. Malo, |
1048 | France, September 1992, Yves Bekkers and Jacques Cohen, eds. Springer, |
1049 | 1992, ISBN 3540-55940-X (real-life thread applications). |
1050 | |
5e549d84 |
1051 | Artur Bergman, "Where Wizards Fear To Tread", June 11, 2002, |
1052 | L<http://www.perl.com/pub/a/2002/06/11/threads.html> |
1053 | |
c975c451 |
1054 | =head1 Acknowledgements |
1055 | |
1056 | Thanks (in no particular order) to Chaim Frenkel, Steve Fink, Gurusamy |
1057 | Sarathy, Ilya Zakharevich, Benjamin Sugars, Jürgen Christoffel, Joshua |
1058 | Pritikin, and Alan Burlison, for their help in reality-checking and |
1059 | polishing this article. Big thanks to Tom Christiansen for his rewrite |
1060 | of the prime number generator. |
1061 | |
1062 | =head1 AUTHOR |
1063 | |
9316ed2f |
1064 | Dan Sugalski E<lt>dan@sidhe.org<gt> |
c975c451 |
1065 | |
1066 | Slightly modified by Arthur Bergman to fit the new thread model/module. |
1067 | |
1068 | =head1 Copyrights |
1069 | |
bfce6503 |
1070 | The original version of this article originally appeared in The Perl |
1071 | Journal #10, and is copyright 1998 The Perl Journal. It appears courtesy |
1072 | of Jon Orwant and The Perl Journal. This document may be distributed |
1073 | under the same terms as Perl itself. |
2605996a |
1074 | |
53d7eaa8 |
1075 | For more information please see L<threads> and L<threads::shared>. |
2605996a |
1076 | |